Abstract:Wearable devices are paramount in health monitoring applications since they provide contextual information to identify and recognize human activities. Although sensor-based human activity recognition (HAR) has been thoroughly examined, prior studies have yet to definitively differentiate between symmetric and asymmetric motions. Determining these movement patterns might provide a more profound understanding of assessing physical activity. The main objective of this research is to investigate the use of wearable motion sensors and deep convolutional neural networks in the analysis of symmetric and asymmetric activities. This study provides a new approach for classifying symmetric and asymmetric motions using a deep residual network incorporating channel and spatial convolutional block attention modules (CBAMs). Two publicly accessible benchmark HAR datasets, which consist of inertial measurements obtained from wrist-worn sensors, are used to assess the model's efficacy. The model we have presented is subjected to thorough examination and demonstrates exceptional accuracy on both datasets. The ablation experiment examination also demonstrates noteworthy contributions from the residual mappings and CBAMs. The significance of recognizing basic movement symmetries in increasing sensor-based activity identification utilizing wearable devices is shown by the enhanced accuracy and F1-score, especially in asymmetric activities. The technique under consideration can provide activity monitoring with enhanced accuracy and detail, offering prospective advantages in diverse domains like customized healthcare, fitness tracking, and rehabilitation progress evaluation.

What problem does this paper attempt to address?

The problem that this paper attempts to solve is: how to accurately identify symmetric and asymmetric human activities using wearable sensors. Specifically, although existing sensor - based human activity recognition (HAR) methods have been widely studied, most of them fail to clearly distinguish between symmetric activities (such as walking, running) and asymmetric activities (such as playing golf, throwing a ball). This distinction is crucial for evaluating the nature of physical activities and for applications in fields such as rehabilitation. To achieve this goal, the author proposes an innovative method that combines the Deep Residual Network and the Convolutional Block Attention Module (CBAM) to improve the classification accuracy of symmetric and asymmetric activities. By using two publicly available benchmark datasets (WISDM - HARB and UTwente), this study verifies the effectiveness of the proposed model and demonstrates its superior performance in identifying asymmetric activities. ### Specific problem description 1. **Distinguishing between symmetric and asymmetric activities**: - Symmetric activities: refer to actions where the movements on both sides of the body are mirror - symmetric, such as walking, running, etc. - Asymmetric activities: refer to actions that lack symmetry or balance, such as typing, eating, etc. 2. **Limitations of existing methods**: - Most existing HAR methods fail to effectively distinguish between symmetric and asymmetric activities. - This distinction is very important for understanding movement patterns and their potential health implications, especially in the field of rehabilitation, where asymmetric activities may be indicators of certain diseases. 3. **Proposed new method**: - This study introduces a new deep - learning model (CNN - ResBLSTM - CBAM), which combines the Convolutional Neural Network (CNN), the Residual Network (ResNet), and the attention mechanism (CBAM). - By using CBAM, the model can more effectively capture and process symmetric and asymmetric features in sensor data. 4. **Experimental verification**: - Experiments are carried out using two publicly available benchmark datasets (WISDM - HARB and UTwente). - The experimental results show that the proposed model performs well in identifying symmetric and asymmetric activities, and is superior to traditional methods especially in the identification of asymmetric activities. ### Summary The main contribution of this paper is the proposal of a new deep - learning framework that can effectively distinguish between symmetric and asymmetric human activities in data based on wearable sensors. This not only improves the accuracy of activity recognition, but also provides more detailed monitoring means for fields such as rehabilitation and fitness tracking.

Deep Residual Network with a CBAM Mechanism for the Recognition of Symmetric and Asymmetric Human Activity Using Wearable Sensors

Hybrid convolution neural network with channel attention mechanism for sensor-based human activity recognition

Deep Wavelet Convolutional Neural Networks for Multimodal Human Activity Recognition Using Wearable Inertial Sensors

A Multi-dimensional Parallel Convolutional Connected Network Based on Multi-source and Multi-modal Sensor Data for Human Activity Recognition

A Multidimensional Parallel Convolutional Connected Network Based on Multisource and Multimodal Sensor Data for Human Activity Recognition

Deep Convolutional Neural Network with RNNs for Complex Activity Recognition Using Wrist-Worn Wearable Sensor Data

HARMamba: Efficient and Lightweight Wearable Sensor Human Activity Recognition Based on Bidirectional Mamba

Deep CNN-LSTM With Self-Attention Model for Human Activity Recognition Using Wearable Sensor

Bimodal HAR-An Efficient Approach to Human Activity Analysis and Recognition Using Bimodal Hybrid Classifiers.

On the Use of a Convolutional Block Attention Module in Deep Learning-Based Human Activity Recognition with Motion Sensors

Deep, Convolutional, and Recurrent Models for Human Activity Recognition using Wearables

WMNN: Wearables-Based Multi-Column Neural Network for Human Activity Recognition

Decoding Human Activities: Analyzing Wearable Accelerometer and Gyroscope Data for Activity Recognition

Wearable Sensor-Based Residual Multifeature Fusion Shrinkage Networks for Human Activity Recognition

Human Activity Recognition Based on Wireless Electrocardiogram and Inertial Sensors

Recognizing human activities with the use of Convolutional Block Attention Module

Human Activity Recognition with Convolutional Neural Netowrks

Sensor-Based Human Activity Recognition with Spatio-Temporal Deep Learning

Deep Residual Bidir-LSTM for Human Activity Recognition Using Wearable Sensors

Wearable sensors based on artificial intelligence models for human activity recognition

A Novel CNN-BiLSTM-GRU Hybrid Deep Learning Model for Human Activity Recognition